Dashboard Demo — Subtle Theft (10% Skimming)
This simulation shows how MineGuard handles subtle, harder-to-detect theft. Days 1-10 establish a normal baseline. On Day 11, a worker begins diverting only ~10% of gold production. At 10%, the theft signal overlaps significantly with normal day-to-day variance (±5%), making detection much slower.
20% Skim (main demo)
Clear signal. 95% confidence by Day 26. Red alert triggered.
10% Skim (this page)
Subtle signal — overlaps with noise. ~75% confidence by Day 30. Needs more time.
The key insight: even 10% theft is detectable — the system sees the pattern emerging, it just needs more days of data to separate signal from noise. Without the daily assay, this level of theft would be invisible forever.
Equipment Runtime (Hours/Day)
Fuel Verification — Expected vs Actual
Fuel consumption closely matches runtime predictions, ruling out equipment issues.
Daily Electrochemical Assay — Ground Richness
Every day, workers scoop ~20g of raw ground into a jar with bleach+HCl every 30 minutes. Gold dissolves passively throughout the work day. At end of day, an SWV measurement gives the actual ground richness in g/t — the worker cannot fake this number.
Why this matters: Without the daily assay, a low gold cleanout could mean “bad ground” or “skimming” — you can't tell which. The assay independently verifies ground quality, so when the assay says the ground is rich but cleanout is low, the only explanation is human intervention. Notice how assay grades (green) vary naturally while cleanout-implied grades (blue) diverge downward after Day 11.
Recovery Ratio — Cleanout vs Assay Prediction
The ratio of actual cleanout to assay-predicted gold should stay near 100%. After Day 11 it drops to ~80% — the assay says the ground is good, but gold isn't showing up at cleanout.
Yield Prediction Formula
MineGuard uses today's assay grade (not a fixed assumption) to calculate expected gold. The daily electrochemical assay provides the actual ground richness, making the prediction accurate even when ground quality varies.
Expected Gold (g) = Material (tons) × Daily Assay Grade (g/t) × Recovery Rate
Example — Day 5 (Feb 5)
28.3 g = 46 × 0.88 × 0.7
The assay grade changes every day because ground quality in Guyana is highly variable (some days 0.7 g/t, some days 1.3 g/t). The daily assay captures this reality. The recovery rate (70%) is established during the supervised baseline period and stays relatively constant for the same equipment.
Yield Comparison — Assay-Predicted vs Cleanout (grams)
“Predicted” is calculated from the daily assay grade — not a fixed number. When assay says the ground is rich but cleanout is low, gold is going missing.
Statistical Confidence — Anomaly Detection
Confidence builds over time as anomalous data accumulates. The 95% threshold marks detection certainty.
Alert Timeline
Cross-correlation log: when fuel matches runtime and the daily assay confirms good ground, but cleanout gold is consistently low — the only explanation is human intervention.
Gold Yield Calculator
Enter your own values to see how the daily assay enables discrepancy detection. The “Assay Grade” is what the electrochemical measurement says the ground contains — the worker cannot fake this.
Input Parameters
Calculation Results
Cleanout is 14.3% below the assay prediction. Worth monitoring over the next few days.